Epistasis II

Cell and developmental biologists have a different working definition of what epistasis is from what I described in the previous post.  In these fields, multigene pathways whose normal function is required for a particular phenotype are studied. Two genes show epistasis if the double mutant phenotype is either (a) the same or not much worse than either single mutant phenotype, (b) similar to or the same as wild type (a compensatory interaction), or (c) worse than the sum of the single mutant effects.
(a) can occur in a linear molecular pathway [e.g. compound a is converted by enzyme A to compound b, which is then converted by enzyme B to compound c, which is then converted by enzyme C to compound d].  In an individual with a mutation in the gene enconding enzyme B, there is not much production of compound d.  Likewise, an individual with a mutation in the gene encoding enzyme C also does not product much of the compound d.  In an individual with both mutations, there is again very little production of compound d.  Thus the presence of one of the mutants masks the effects of the other mutant in the same genetic background.  Biochemical methods that quantify the amounts of all of the compounds in the pathway (a,b,c, and d) can be used to identify which steps(genes) are effected in each mutant.  This type of biochemical genetic work in bacteria and fungi greatly helped to initially characterize most of the pathways of basic cellular metabolism in the middle part of the 20th century.

Two pathways may act in parallel, such that when one is knocked out by a mutation, the other is able to mostly or completely compensate.  In this situation, the first mutation may be nearly or completely undetectable, or it may cause only a modest decrease in the function of its pathway and hence a mildly deleterious phenotype.  The second mutation would knock out  the parallel "backup" pathway, resulting in a great or completely loss of the product of the two pathways, and hence a severe phenotype or lethality (a dead organism).  This is the basis of many "enhancer suppressor screens" in model organisms (predominantly Drosophila and Caenorhabditis in animals) that have been used with high effectiveness to characterize developmental and cellular signaling pathways.

The parallel pathway situation can also be the basis of compensatory epistasis..  If pathway 1 and pathway 2 both contribute to the production of compound A and a certain amount or concentration of A is needed (i.e. not too much and not too little), then a mutation that increases the production of compound A from pathway 1 can be compensated by a second mutation that decreases the production of A from pathway 2.  Of course, many parallel pathways likely have this type of compensatory action built into them.  A could act to inhibit gene expression or protein/enzyme activity in either or both of the pathway, such that an excess of A acts to turn down its own production until its level falls below a certain threshold.